model_cards/article.md

# Model documentation & parameters

## Parameters

### Property
The supported properties are:
  - `Metal NonMetal Classifier`: Classifying whether a crystal could be metal or nonmetal using a [RandomForest classifier](https://www.nature.com/articles/s41524-022-00850-3)
  - `Metal Semiconductor Classifier`: Classifying whether a crystal could be metal or semiconductor using the [CGCNN framework](https://link.aps.org/doi/10.1103/PhysRevLett.120.145301).
  - `Poisson Ratio`:  Predicted using the [CGCNN framework](https://link.aps.org/doi/10.1103/PhysRevLett.120.145301).
  - `Shear Moduli`:  Predicted using the [CGCNN framework](https://link.aps.org/doi/10.1103/PhysRevLett.120.145301).
  - `Bulk Moduli`:  Predicted using the [CGCNN framework](https://link.aps.org/doi/10.1103/PhysRevLett.120.145301).
  - `Fermi Energy`:  Predicted using the [CGCNN framework](https://link.aps.org/doi/10.1103/PhysRevLett.120.145301).
  - `Band Gap`:  Predicted using the [CGCNN framework](https://link.aps.org/doi/10.1103/PhysRevLett.120.145301).
  - `Absolute Energy`:  Predicted using the [CGCNN framework](https://link.aps.org/doi/10.1103/PhysRevLett.120.145301).
  - `Formation Energy`:  Predicted using the [CGCNN framework](https://link.aps.org/doi/10.1103/PhysRevLett.120.145301).


### Input file for crystal model
The file with information about the metal. Dependent on the property you want to predict, the format of the file differs:
- `Metal NonMetal Classifier`. It requires a single `.csv` file with the metal (chemical formula) in the first column and the crystal system in the second.
- **All others**: Predicted with CGCNN. The input can either be a single `.cif` file (to predict a single metal) or a `.zip` folder which contains multiple `.cif` (for batch prediction)



# Model card - CGCNN

**Model Details**: Eight CGCNN models trained to predict various properties for crystals. 

**Developers**: [CGCNN's](https://github.com/txie-93/cgcnn) developers.

**Distributors**: Original authors' code wrapped and distributed by GT4SD Team (2023) from IBM Research.

**Model date**: 2018.

**Algorithm version**: Models trained and distributed by the original authors.
- **Metal Semiconductor Classifier**: Model trained to classify whether a crystal could be metal or semiconductor using instances from this [database](https://aip.scitation.org/doi/10.1063/1.4812323) that includes a diverse set of inorganic crystals ranging from simple metals to complex minerals..
- **Poisson Ratio**: Model to predict the Poisson ratio trained on 2041 instances from this [database](https://aip.scitation.org/doi/10.1063/1.4812323) that includes a diverse set of inorganic crystals ranging from simple metals to complex minerals.
- **Shear Moduli**: Model to predict the Shear moduli trained on 2041 instances from this [database](https://aip.scitation.org/doi/10.1063/1.4812323) that includes a diverse set of inorganic crystals ranging from simple metals to complex minerals. Unit log(GPa).
- **Bulk Moduli**: Model to predict the Bulk moduli trained on 2041 instances from this [database](https://aip.scitation.org/doi/10.1063/1.4812323) that includes a diverse set of inorganic crystals ranging from simple metals to complex minerals. Unit log(GPa).
- **Fermi Energy**: Model to predict the Fermi energy trained on 28046 instances from this [database](https://aip.scitation.org/doi/10.1063/1.4812323) that includes a diverse set of inorganic crystals ranging from simple metals to complex minerals. Unit eV.
- **Band Gap**: Model to predict the Band Gap trained on 16458 instances from this [database](https://aip.scitation.org/doi/10.1063/1.4812323) that includes a diverse set of inorganic crystals ranging from simple metals to complex minerals. Unit eV.
- **Absolute Energy**: Model to predict the Absolute energy trained on 28046 instances from this [database](https://aip.scitation.org/doi/10.1063/1.4812323) that includes a diverse set of inorganic crystals ranging from simple metals to complex minerals. Unit eV/atom.
- **Formation Energy**: Model to predict the formation energy trained on 28046 instances from this [database](https://aip.scitation.org/doi/10.1063/1.4812323) that includes a diverse set of inorganic crystals ranging from simple metals to complex minerals. Unit eV/atom.
  
**Model type**: Crystal Graph Convolutional Neural Networks (CGCNN) that take an arbitary crystal structure to predict material properties.

**Information about training algorithms, parameters, fairness constraints or other applied approaches, and features**: 
See the [CGCNN](https://link.aps.org/doi/10.1103/PhysRevLett.120.145301) paper for details.

**Paper or other resource for more information**: 
The [CGCNN](https://link.aps.org/doi/10.1103/PhysRevLett.120.145301) paper. See the [source code](https://github.com/txie-93/cgcnn) for details.

**License**: MIT

**Where to send questions or comments about the model**: Open an issue on [CGCNN](https://github.com/txie-93/cgcnn) repo.

**Intended Use. Use cases that were envisioned during development**: Materials research.

**Primary intended uses/users**: Researchers using the model for model comparison or research exploration purposes.

**Out-of-scope use cases**: Production-level inference, producing molecules with harmful properties.

**Factors**: N.A.

**Metrics**: N.A.

**Datasets**: Different ones, as described under **Algorithm version**.

**Ethical Considerations**: No specific considerations as no private/personal data is involved. Please consult with the authors in case of questions.

**Caveats and Recommendations**: Please consult with original authors in case of questions.

Model card prototype inspired by [Mitchell et al. (2019)](https://dl.acm.org/doi/abs/10.1145/3287560.3287596?casa_token=XD4eHiE2cRUAAAAA:NL11gMa1hGPOUKTAbtXnbVQBDBbjxwcjGECF_i-WC_3g1aBgU1Hbz_f2b4kI_m1in-w__1ztGeHnwHs)


# Model card - RandomForestMetalClassifier

ToDo...


# Citation

```bib
@article{PhysRevLett.120.145301,
  title = {Crystal Graph Convolutional Neural Networks for an Accurate and Interpretable Prediction of Material Properties},
  author = {Xie, Tian and Grossman, Jeffrey C.},
  journal = {Phys. Rev. Lett.},
  volume = {120},
  issue = {14},
  pages = {145301},
  numpages = {6},
  year = {2018},
  month = {Apr},
  publisher = {American Physical Society},
  doi = {10.1103/PhysRevLett.120.145301},
  url = {https://link.aps.org/doi/10.1103/PhysRevLett.120.145301}
}
```